Scan barcode
A review by sbenzell
The Theory of Moral Sentiments by Adam Smith
4.0
Honestly extremely boring for large swathes. People have emphathy. This makes them like people who are generous and not like people who are jerks. They like it when generous people are rewarded and jerks are punished. What makes something generous or jerky? When it is much nicer -on the one hand- or selfisher/stupider - on the other hand - than what is socially expected/required (I think there is proto-version of utilitarianism in here).
What's begged? Well, why should a certain act be perceived as kind/generous or jerky? And why should we base our moral system on some sort of majority vote? On the one hand Smith does accept that custom will move around what is considered kind/generous, and that because of correlations, sometimes vices will be percieved as virtues because they are associated with strong or lauded people. It seems like this should fly off into a pure post-modern "what is good is what is percieved as what is good", but Smith thinks that the most basic human desires -- for social approval ("approbation"), wealth ("convenience"), or other "selfish passions" are universal enough, when combined with empathy, to keep morality from flying completely off the rails.
As for "the opinion of an impartial spectator" being normative, this is an intriguing idea that Smith gets credit for introducing. Again we are faced with the question of exactly how impartial we want the spectator to be. A complete alien might not understand our customs well enough to understand how actions were intended, but a neighbor might have a selfish or partisan interest. This seems like asking for a view "behind the veil of ignorance" a move that has been shown to have lots of problems (how do you have a view before you have an identity?).
Still, while this isn't an entirely satisfactory normative theory of morality, it is an intriguing and powerful positive theory of morality -- that moral individuals build a model in their heads of what an impartial observer would think about the situation, and try to deport themselves in that matter. This relates to the idea of the super-ego from Freud, and also to the idea of Reinforcement Learning for LLM models (in order to succeed in RL-ing, the LLM will likely need a good model of what we want from the models, and if trained by an army of, well, representative internet users, this sounds a lot like the 'impartial observer'/view from nowhere which is the idea of Smith).
But gosh so boring and normie! So reasonable! There's some claims I could quibble with but my main take away is how modern and uncontroversial some of this feels. Part of it seems like society-as-it-is-moral-apologetics ala Burkean conservatism, but it seems edgy for the time (e.g. by being irreligious) in a way which is forward looking.
I'll try to find a better book for us to read next time @AndreyFradkin!
What's begged? Well, why should a certain act be perceived as kind/generous or jerky? And why should we base our moral system on some sort of majority vote? On the one hand Smith does accept that custom will move around what is considered kind/generous, and that because of correlations, sometimes vices will be percieved as virtues because they are associated with strong or lauded people. It seems like this should fly off into a pure post-modern "what is good is what is percieved as what is good", but Smith thinks that the most basic human desires -- for social approval ("approbation"), wealth ("convenience"), or other "selfish passions" are universal enough, when combined with empathy, to keep morality from flying completely off the rails.
As for "the opinion of an impartial spectator" being normative, this is an intriguing idea that Smith gets credit for introducing. Again we are faced with the question of exactly how impartial we want the spectator to be. A complete alien might not understand our customs well enough to understand how actions were intended, but a neighbor might have a selfish or partisan interest. This seems like asking for a view "behind the veil of ignorance" a move that has been shown to have lots of problems (how do you have a view before you have an identity?).
Still, while this isn't an entirely satisfactory normative theory of morality, it is an intriguing and powerful positive theory of morality -- that moral individuals build a model in their heads of what an impartial observer would think about the situation, and try to deport themselves in that matter. This relates to the idea of the super-ego from Freud, and also to the idea of Reinforcement Learning for LLM models (in order to succeed in RL-ing, the LLM will likely need a good model of what we want from the models, and if trained by an army of, well, representative internet users, this sounds a lot like the 'impartial observer'/view from nowhere which is the idea of Smith).
But gosh so boring and normie! So reasonable! There's some claims I could quibble with but my main take away is how modern and uncontroversial some of this feels. Part of it seems like society-as-it-is-moral-apologetics ala Burkean conservatism, but it seems edgy for the time (e.g. by being irreligious) in a way which is forward looking.
I'll try to find a better book for us to read next time @AndreyFradkin!